The Effect of Topical Bevacizumab on Corneal Neovascularization

Sang Woo Kim, MD,1 Byung Jin Ha, MD,1 Eung Kweon Kim, MD, PhD,1,2 Hungwon Tchah, MD,3 Tae-im Kim, MD1,2 Purpose: To examine the effect of topical bevacizumab on corneal neovascularization (NV) over a period of 3 months. Design: Prospective, nonrandomized, masked observational case series. Participants: Ten eyes of 7 patients with corneal NV. Methods: Patients received topical bevacizumab (1.25%) twice daily. Ophthalmic evaluations included visual acuity, slit-lamp examination, and tonometry. Main Outcome Measures: Corneal NV and changes in ophthalmic evaluations. Results: Decreased corneal NV was noted in 7 of 10 eyes, usually within 1 month of treatment. Epitheli- opathy (epithelial defect, epithelial erosion) was observed in 6 of 10 eyes, 1 resulting in corneal thinning. Adverse effects generally appeared during the second month of treatment. Conclusions: Topical application of bevacizumab was effective in reducing corneal NV within the ﬁrst month. However, by the second month there was an increased risk of adverse effects. Ophthalmology 2008;115: e33–e38 © 2008 by the American Academy of Ophthalmology.

Recent advances in understanding the molecular mechanisms malignancies. A promising area of clinical research is the underlying angiogenesis have facilitated the development of combination of bevacizumab with cytotoxic chemotherapy for promising new therapies for neovascular ocular diseases.1,2 colorectal cancer.8 Bevacizumab was granted approval for use Corneal neovascularization (NV) is a challenging condition, as first-line therapy in metastatic colorectal cancer in combi- and because corneal clarity and avascularity are critical for nation with fluorouracil-based chemotherapy. With its unique maintaining vision, developing treatments for corneal NV is antineoplastic activity and tolerability, bevacizumab provides crucial. therapeutic benefits for cancer patients by improving overall Corneal NV occurs as a result of disequilibrium between response rates, time to progression, and survival. angiogenic and antiangiogenic stimuli.3,4 Corneal angiogenic Bevacizumab has also been a valuable addition to the factors include vascular endothelial growth factor (VEGF),2,5 limited treatment options available for age-related macular with studies showing that VEGF activation can induce corneal degeneration. Encouraging results for intravitreal bevacizumab NV, and that inhibition of VEGF can block new vessel for- treatment for age-related macular degeneration have led to mation in human and animal cornea model.6,7 Bevacizumab trials involving macular edema in central vein occlusion, iris (Avastin, Genentech, San Francisco, CA) is a full-length hu- NV, and subsequently, severe proliferative diabetic retinopa- manized murine monoclonal antibody directed against VEGF. thy.9–11 Such trials have encouraged investigations on the use Bevacizumab is a powerful anti-NV agent used for cancer of topical bevacizumab for corneal NV. Topically adminis- therapy. There are numerous ongoing clinical studies examin- tered bevacizumab was found to inhibit corneal NV after ing the effect of bevacizumab in various solid and hematologic chemical injury in an experimental rat model.12 In humans, a short-term follow-up study showed that topical bevacizumab 13 Originally received: September 16, 2007. reduced corneal NV in patients with significant corneal NV. Final revision: January 24, 2008. Systemic administration of bevacizumab is reported to Accepted: February 15, 2008. have a low incidence of adverse effects such as hyperten- Available online: April 23, 2008. Manuscript no. 2007-1206. sion and thrombosis,14 and a recent report suggested that 1 Corneal Dystrophy Research Institute, Department of Ophthalmology, small doses delivered topically would not produce serious Yonsei University College of Medicine, Seoul, Republic of Korea. systemic effects.15 Reported ocular adverse events after 2 BK21 Project Team of Nanobiomaterials for Cell-based Implants, Yonsei intravitreal bevacizumab injections include mild anterior University, Seoul, Republic of Korea. 16 3 chamber inflammation, subconjunctival hemorrhage, mild Department of Ophthalmology, Asan Medical Center, University of Ulsan, 17 Seoul, Republic of Korea. vitritis, corneal abrasion, and retinal pigment epithelial detachment.1 There are few reports regarding the effect of The authors have no conflicts of interest related to the article. topical bevacizumab administration on corneal wound healing. This study was supported by Korean Research Foundation grant 2006- KRF-531-E00121. The present study examined the effect of topical bevaci- Correspondence to Tae-im Kim, MD, Severance Hospital, Yonsei Univer- zumab on corneal NV of various etiologies. In general, the sity College of Medicine, Department of Ophthalmology, #250 Sungsanno, treatment was found to reduce corneal NV. However, de- Seodaemun-gu, 120-752, Seoul, Republic ofE-mail: Korea. [email protected]. layed side effects appeared in several patients.

Table 1. Summary of Cases of Topical Bevacizumab Treatments

Age (yrs)/ Medications Other Patient Gender Eye Diagnosis than Bevacizumab Comments 1 31/M Right Severe acid injury Artificial tears, levofloxacin, Improvement of NV 0.125% prednisolone acetate Epithelial defect followed by stromal thinning and descemetocele 2 64/F Both Stevens-Johnson syndrome Artificial tears, levofloxacin, Improvement of NV on both eyes 0.125% prednisolone acetate Epithelial defect on right eye 3 18/F Both Stevens-Johnson syndrome Artificial tears Improvement of NV Epithelial erosion on both eyes 4 24/M Right Herpetic keratitis Artificial tears No improvement of NV 5 39/M Right Penetrating keratoplasty Artificial tears, levofloxacin, Improvement of NV 0.125% prednisolone acetate Epithelial defect 6 24/M Both Stevens-Johnson syndrome Ofloxacine, 0.1% fluorometholone No improvement of NV on either eye 7 54/F Right Pterygium resection Artificial tears, ofloxacin, Cessation of NV development 0.125% prednisolone acetate Reduction of opacity

F ϭ female; M ϭ male; NV ϭ neovascularization.

Patients and Methods not fully regained its normal thickness even 3 months after cessation of bevacizumab treatment (Fig 1D). There was a stromal defect A sample of consecutive, eligible patients was identified for in- apparent on slit-lamp examination. clusion in the study over 1 year from 2006 to 2007. Patients with corneal NV were treated with 1.25% topical bevacizumab (12.5 Patient 2 mg bevacizumab with 1 mL normal saline was prepared by Sev- erance Hospital pharmacy) twice daily for 3 months. All patients A 64-year-old woman with Stevens-Johnson syndrome presented were 18–64 years of age and continued to use other topical with ocular surface inflammation and corneal NV. Topical cortico- medications (Table 1). Patients with infectious keratitis or other steroid treatment for several months had little effect. Examination at epithelial disease were excluded. Data collected for analysis in- this stage showed diffuse scarring, superficial and stromal corneal cluded best-corrected visual acuity, slit-lamp photography, and NV, and conjunctival injection, but corneal epithelial integrity was tonometry. This study was approved by the Institutional Review well maintained in both eyes (Fig 1E) and corneal conjunctivalization Board of Severance Hospital. All patients were informed of the was not observed. effects of topical bevacizumab and their consent was obtained. The patient was prescribed topical bevacizumab for both eyes to reduce the corneal NV. After 1 month, the superficial and deep stromal corneal NV had reduced significantly. However, a spontane- Results ous epithelial defect subsequently developed on her right eye after 6 weeks (Fig 1F). Bevacizumab treatment was discontinued, and the Patient 1 epithelial defect healed 1 week later. A 31-year-old man who had a grade IV acid injury, from the classification of Dua et al,18 to the right eye 4 years ago underwent Patient 3 symblepharon lysis and conjunctival sac reconstruction 4 times, and limbal cell transplantation twice. However, the corneal NV did not An 18 year-old-girl with Stevens-Johnson syndrome complained of respond to consecutive surgical treatments combined with topical decreased vision and showed corneal NV and opacity in both eyes. steroid therapy. The best-corrected visual acuity in the right eye was Topical corticosteroid therapy for several months had little effect on finger counting at 30 cm. Slit-lamp examination revealed diffuse reducing the diffuse corneal opacity or the superficial and stromal superficial and deep stromal corneal NV with significant corneal corneal NV (Fig 2A). Topical bevacizumab was then prescribed. One scarring and some conjunctival symblepharons (Fig 1A). month later, there was a reduction in NV, and visual acuity improved Topical bevacizumab was prescribed to improve the prognosis of from 20/40 to 20/25 in the left eye (Fig 2B). However, diffuse corneal subsequent limbal and/or corneal transplantation. The eye was treated erosion was observed on both eyes and topical bevacizumab treatment with topical 1.25% bevacizumab twice daily, in addition to artificial was ceased. Two weeks later the corneal erosion had resolved. tears and a topical steroid (0.125% prednisolone acetate, Samil Phar- maceutics, Seoul, Republic of Korea). Patient 4 The patient was assessed using slit-lamp examination and tonometry. After 4 weeks of bevacizumab treatment, the superficial and A 24-year-old man with herpetic stromal keratitis had a single large deep stromal corneal NV was markedly reduced. Intraocular pressure conjunctival vessel emerging from the 4-o’clock position at the lim- remained at baseline levels during treatment. Three months after bus and passing over the corneal center in the right eye (Fig 2C). treatment commenced, a spontaneous epithelial defect appeared in The opacity resulted in a best-corrected visual acuity of 20/30. central cornea (Fig 1B). Topical bevacizumab and steroid treatment Topical bevacizumab treatment was commenced, and 2 months were discontinued, a bandage contact lens was applied, and levofloxa- later the vascular caliber had slightly decreased and opacity had cin eye drops were administered (Fig 1C). However, healing of this reduced. The best-corrected visual acuity was 20/25, and no other epithelial defect was slow, and stromal thinning and descemetocele corneal problems were evident. Topical bevacizumab treatment appeared 1 week later. One month later, a thin layer of stroma had was discontinued, and the improvement in visual acuity was still regenerated above Descemet’s membrane, but the stromal layer had present at a follow-up 5 months later (Fig 2D). e34 Kim et al ⅐ Effect of Topical Bevacizumab

Figure 1. The effect of topical bevacizumab in cases 1 and 2. Case 1 (A–D). A, Before topical bevacizumab treatment. Note the corneal opacity and neovascularization (NV). B, Three months after topical bevacizumab treatment. Note the spontaneous epithelial defect in the central corneal area. C, Stromal thinning and descemetocele development. D, One month after cessation of treatment. Note healing of the epithelial defect and the thin stroma on top of the bare Descemet’s membrane. Case 2 (E, F). E, Before topical bevacizumab treatment. Note the diffuse superﬁcial and deep corneal NV from the conjunctiva passing over the limbus to the center of cornea. F, One month after topical bevacizumab treatment. Note the decrease in central corneal NV, and the appearance of a spontaneous epithelial defect.

Patient 5 Patient 6 A 39-year-old man who had undergone penetrating keratoplasty in A 24-year-old man with Stevens-Johnson syndrome complained of the right eye showed slowly developing corneal NV across the decreased vision and was diagnosed with corneal NV in both eyes graft and host junction (Fig 2E). Despite topical steroid treatment, (Fig 3A). Topical bevacizumab was prescribed. Examinations after corneal NV developed and did not regress. Twice daily application 3 months showed the treatment had no effect on the corneal NV, of topical bevacizumab was prescribed. One month later there was nor did it adversely apparent affect the ocular surface (Fig 3B). evidence of reduced corneal NV (Fig 2F). However, a spontaneous epithelial defect developed in the inferior cornea. A topical anti- Patient 7 biotic was prescribed, topical steroid treatment was discontinued, and bevacizumab treatment was reduced to once daily. After 2 A 54-year-old woman complained of remaining vascularization 1 weeks the epithelial defect remained. Topical bevacizumab was month after surgical pterygium removal (Fig 3C). A conjunctival then discontinued and a pressure patch was placed and changed vessel progressed across the limbus to the wound site. Topical daily. After 1 week the epithelial defect had healed completely. steroid treatment did not prevent vascular growth. Twice daily

e35 Ophthalmology Volume 115, Number 6, June 2008

Figure 2. The effect of topical bevacizumab in cases 3–5. Case 3 (A, B). A, Before topical bevacizumab treatment. Note the severe corneal neovascularization (NV) and dense central corneal opacity. B, One month after topical bevacizumab treatment. Note the reduction in corneal NV and opacity. Case 4 (C, D). C, Before topical bevacizumab treatment. Note the single vessel from the limbus passing over the corneal center in the right eye. D, Two months after topical bevacizumab treatment. Note the slightly decreased caliber of the large vessel and reduction in vascular leakage. Case 5 (E, F). E, Before topical bevacizumab treatment. Note the corneal NV over the donor host junction in the right eye after penetrating keratoplasty. F, One month after topical bevacizumab treatment. Note the delayed progression of corneal NV. topical bevacizumab was prescribed. Two weeks later, topical Newly or already formed corneal NV elevates the risk of steroid treatment was stopped owing to an elevated intraocular subsequent graft rejection after corneal transplantation.19 pressure (up to 25 mmHg) and antiglaucoma medication was Medical and surgical therapies used to reduce corneal NV added. One month later, corneal NV progression had ceased, and include corticosteroids, nonsteroidal anti-inflammatory agents, the inflammatory reaction of the wound had stabilized (Fig 3D). 3 No significant adverse effects were observed. All treatment was laser photocoagulation, and needle diathermy. Many of then ceased, and 3 months later the inflammatory reaction at the these therapies have not only demonstrated limited success ocular surface remained under control. but also have associated adverse effects. Anti-VEGF therapy targeting corneal NV has recently showed successful results in many preliminary in vivo animal experiments. In Discussion rat models, topical bevacizumab (4 mg/mL) applied twice daily for 1 week reduced chemically induced corneal NV,12 Corneal NV can lead to vision loss, is often difficult to and anti-VEGF antibody implanted in neovascularized cor- manage, and may lead to a need for corneal transplantation. neal stroma suppressed corneal NV.7 However, such find- e36 Kim et al ⅐ Effect of Topical Bevacizumab

Figure 3. The effect of topical bevacizumab in cases 6 and 7. Case 6 (A, B). A, Before topical bevacizumab treatment. Note that more than half of the cornea is covered with diffuse superficial and deep corneal neovascularization (NV). B, At 3 months after treatment. Note no significant changes. Case 7(C, D). C, Before topical bevacizumab treatment. Vessels from the conjunctiva crossed from the limbus towards the cornea 1 month after pterygium resection. D, One month after topical bevacizumab treatment. Note no significant vascular change, but reduced corneal opacification. ings only indicate the potential of topical anti-VEGF ther- study showed a stable state of corneal NV and epithelium apy for controlling NV. before treatment. The corneal epithelial defects seen after Vascular endothelial growth factor affects the extracel- use of topical bevacizumab suggest that such treatment may lular matrix metabolism,20 and may regulate structural affect adhesion between epithelium and basement mem- changes associated with vascularization through intact tis- branes, rapidly differentiate the corneal epithelium itself, or sue layers. It upregulates platelet activating factor activity, inhibit or delay normal wound healing processes. The exact which enhances urokinase-type plasminogen activator gene mechanisms need to be elucidated through further studies. 21,22 expression in corneal epithelium. Induced urokinase- Corneal NV can be triggered by hypoxia and wound heal- type plasminogen activator, which is also upregulated by 23 ing. But, it is difficult to know the full consequences of wounding itself, has a role in cell migration, cell adhesion, blocking NV. Anti-VEGF therapy seems effective in sup- and tissue remodeling, and thus plays a key role in initiating 24,25 pressing new vessel formation and vascular leakage, which corneal wound healing and recovery. VEGF also in- can improve visual function. However, blocking VEGF creases the fibrinolytic activity of endothelial cells within function may also disrupt wound healing, causing ischemia fibrin matrices with the involvement of VEGF receptor-2, and increasing overall tissue damage. tissue type plasminogen activator, and matrix metalloproteinases.26 In addition, transfection of the soluble VEGF Anti-VEGF therapy is considered as a possible tool for receptor (sflt-1) gene was shown to attenuate pulmonary controlling NV in many clinical fields. Anti-VEGF agents fibrosis in a mouse model of bleomycin-induced pneumo- are becoming more broadly used for various ocular diseases, pathy, suggesting that an anti-VEGF approach might also including corneal NV after chemical injury, choroidal NV, 10,13,16,28 offer a suitable antifibrotic therapy.27 Thus, VEGF seems to and neovascular glaucoma. However, angiogenesis be a critical modulator of wound healing not only via is exquisitely regulated, and while affecting the balance may controlling angiogenesis but also via regulating wound heal- provide some benefits, there may also be some adverse ing and fibrosis development. complications. As such, long-term studies are critical before In the present study, topical application of bevacizumab considering clinical application. Thus, long-term clinical resulted not only in vascular suppression but also caused results and basic studies are required to provide further spontaneous loss of epithelial integrity and progression of knowledge regarding the possible use of bevacizumab treat- stromal thinning. The patients who were recruited in this ment for corneal diseases.

e37 Ophthalmology Volume 115, Number 6, June 2008

References itreal Bevacizumab Safety Study: using the internet to assess drug safety worldwide. Br J Ophthalmol 2006;90:1344–9. 16. Abraham-Marin ML, Cortes-Luna CF, Alvarez-Rivera G, 1. Lynch SS, Cheng CM. Bevacizumab for neovascular ocular et al. Intravitreal bevacizumab therapy for neovascular age- diseases. Ann Pharmacother 2007;41:614–25. related macular degeneration: a pilot study. Graefes Arch Clin 2. Hosseini H, Nejabat M. A potential therapeutic strategy for Exp Ophthalmol 2007;245:651–5. inhibition of corneal neovascularization with new anti-VEGF 17. Spaide RF, Laud K, Fine HF, et al. Intravitreal bevacizumab agents. Med Hypotheses 2007;68:799–801. treatment of choroidal neovascularization secondary to age- 3. Chang JH, Gabison EE, Kato T, Azar DT. Corneal neovascu- related macular degeneration. Retina 2006;26:383–90. larization. Curr Opin Ophthalmol 2001;12:242–9. 18. Dua HS, King AJ, Joseph A. A new classification of ocular 4. Azar DT. Corneal angiogenic privilege: angiogenic and anti- surface burns. Br J Ophthalmol 2001;85:1379–83. angiogenic factors in corneal avascularity, vasculogenesis, and 19. Volker-Dieben HJ, D’Amaro J, Kok-van Alphen CC. Hierar- wound healing (an American Ophthalmological Society the- chy of prognostic factors for corneal allograft survival. Aust N sis). Trans Am Ophthalmol Soc 2006;104:264–302. Z J Ophthalmol 1987;15:11–8. 5. Gan L, Fagerholm P, Palmblad J. Vascular endothelial growth 20. Wu Y, Zhang Q, Ann DK, et al. Increased vascular endothelial factor (VEGF) and its receptor VEGFR-2 in the regulation of growth factor may account for elevated level of plasminogen corneal neovascularization and wound healing. Acta Ophthal- activator inhibitor-1 via activating ERK1/2 in keloid fibro- mol Scand 2004;82:557–63. blasts. Am J Physiol Cell Physiol 2004;286:C905–12. 6. Phillips GD, Stone AM, Jones BD, et al. Vascular endothelial 21. Bernatchez PN, Allen BG, Gélinas DS, et al. Regulation of growth factor (rhVEGF 165) stimulates direct angiogenesis in VEGF-induced endothelial cell PAF synthesis: role of p42/44 the rabbit cornea. In Vivo 1994;8:961–5. 7. Amano S, Rohan R, Kuroki M, et al. Requirement for vascular MAPK, p38 MAPK and PI3K pathways. Br J Pharmacol endothelial growth factor in wound- and inflammation-related 2001;134:1253–62. corneal neovascularization. Invest Ophthalmol Vis Sci 1998; 22. Tao Y, Bazan HE, Bazan NG. Platelet-activating factor en- 39:18–22. hances urokinase-type plasminogen activator gene expression 8. Zondor SD, Medina PJ. Bevacizumab: an angiogenesis inhib- in corneal epithelium. Invest Ophthalmol Vis Sci 1996;37: itor with efficacy in colorectal and other malignancies. Ann 2037–46. Pharmacother 2004;38:1258–64. 23. Watanabe M, Yano W, Kondo S, et al. Up-regulation of 9. Rosenfeld PJ, Fung AE, Puliafito CA. Optical coherence to- urokinase-type plasminogen activator in corneal epithelial mography findings after an intravitreal injection of bevaci- cells induced by wounding. Invest Ophthalmol Vis Sci 2003; zumab (Avastin) for macular edema from central retinal vein 44:3332–8. occlusion. Ophthalmic Surg Lasers Imaging 2005;36:336–9. 24. Irigoyen JP, Muñoz-Cánoves P, Montero L, et al. The plas- 10. Bock F, König Y, Dietrich T, et al. Inhibition of angiogenesis minogen activator system: biology and regulation. Cell Mol in the anterior chamber of the eye [in German]. Ophthalmo- Life Sci 1999;56:104–32. loge 2007;104:336–44. 25. Blasi F. Proteolysis, cell adhesion, chemotaxis, and invasive- 11. Avery RL, Pearlman J, Pieramici DJ, et al. Intravitreal bev- ness are regulated by the u-PA-u-PAR-PAI-1 system. Thromb acizumab (Avastin) in the treatment of proliferative diabetic Haemost 1999;82:298–304. retinopathy. Ophthalmology 2006;113:1695–705. 26. Ratel D, Mihoubi S, Beaulieu E, et al. VEGF increases the 12. Manzano RP, Peyman GA, Khan P, et al. Inhibition of exper- fibrinolytic activity of endothelial cells within fibrin matrices: imental corneal neovascularisation by bevacizumab (Avastin). involvement of VEGFR-2, tissue type plasminogen activator Br J Ophthalmol 2007;91:804–7. and matrix metalloproteinases. Thromb Res 2007;121:203–12. 13. DeStafeno JJ, Kim T. Topical bevacizumab therapy for cor- 27. Chaudhary NI, Roth GJ, Hilberg F, et al. Inhibition of PDGF, neal neovascularization. Arch Ophthalmol 2007;125:834–6. VEGF and FGF signalling attenuates fibrosis. Eur Respir J 14. Fernando NH, Hurwitz HI. Targeted therapy of colorectal 2007;29:976–85. cancer: clinical experience with bevacizumab. Oncologist 28. Zhang SX, Ma JX. Ocular neovascularization: implication of 2004;9(suppl):11–8. endogenous angiogenic inhibitors and potential therapy. Prog 15. Fung AE, Rosenfeld PJ, Reichel E. The International Intrav- Retin Eye Res 2007;26:1–37.

e38